A set of similar images is often created when a person takes multiple photographs of a scene. The set of similar images may have different viewing angles, zoom settings, lighting, and exposure settings. Images similar to a particular image may be available via the internet. For example, images taken at a famous vacation landmark or tourist location may be readily available via the internet.
When multiple images of an object or scene are available, a user may have interest in determining which of the images has the “best” image quality. Traditional methods for image quality evaluation include “full-reference” methods, “reduced reference” methods, and “no reference” methods. “Full reference” methods compare a full first image to a full second image. “Reduced reference” methods compare a full first image to a portion of a full second image. “No reference” methods do not use a second image. Rather, in “no reference” methods, blurriness, blockiness (e.g., due to compression), or noisiness (e.g., due to low light image capture) of a full image is estimated.